Machine learning in trading: theory, models, practice and algo-trading - page 1448

New comment
 
Oleg avtomat:

On the subject of intellectual honesty:

Uncle, aren't you confused? This is a science topic, not a tractor driver's locker room

There is no science in Russia now, what was in the USSR - no one is interested in it for a long time
 
Maxim Dmitrievsky:

Uncle, aren't you confused? This is a science topic, not a tractor driver's shop

There is no science in Russia now, what was in the USSR - no one is interested in it anymore

stupid...

 
Ivan Butko:

Right above your post just above you posted a promotional video of a neural networker. Pay attention

I took a look at the video.
There really are screenshots like yours.
Apparently, the developer is the same and made different products on the same engine, but didn't put yours together with others on his site, so I thought it was a switch. I apologize.

It's not clear why it's not on the developer's site. Has it been withdrawn from sale or what?
 
Again I have a question what is fed to the input. I understood that it is formed in some magical way only on the basis of the loaded kotir. whatever it is, we need repeated tests that it would be possible to make at least some preliminary conclusions.
 
Mihail Marchukajtes:
My question is again what is the input. I think it is formed in some magic way only on the basis of a loaded quote.

This is the most probable that it has its own predictors, as I understood the trading signals go through the same software, so the buyer does not see what's going on inside. However, such secrecy may contain fraud, so I would teach on quotes cut off in the terminal, and after training I would add quotes and only then you can be more confident that the product is not peeking during training.

 
Aleksey Vyazmikin:

This is the most probable that it has its own predictors, as I understood the trading signals go through the same software, so the buyer does not see what's going on inside. The only way to be sure that the product is not peeping into the training is to add quotes after the training.

What predictors there can be that just based on a quote itself they can make similar forecasts. The internal architecture alone won't make the weather, so such a good result is highly questionable. The point of NS is not only to get a good model and repeat this result stably, obtained singularly good eqivity as a rule is an accident and has nothing to do with any intellect. The strange thing is that there is no possibility to load inputs or at least to be able to choose. I take information from exchange by making all sorts of conversions and I can't brag about the result, and here I just use quotes and get the same result.... I do not believe ...
 
Elibrarius:

Google to help.
From the screenshot I found https://megatrader.org/ru/primery-i-obzor-programmy.


The screenshot is identical to yours. It costs 24900 rubles.

Is it?

What are these expenses for, gentlemen? There is also NeuroPro by comrade Tsaregorodtsev, and if you want to pay, then good old NeuroShell Trader.

 
The Grail:

What are these expenses for, gentlemen? There is also NeuroPro by comrade Tsaregorodtsev, and if you want to pay, then good old NeuroShell Trader.

I'm not going to pay - if there are good predictors, any network/forest will handle them. Including the alglibs or something from R.

From settings you can see that it takes 8 bars M15 for entry.

 
Mihail Marchukajtes:
Again I have a question what is fed to the input. I think that it is formed in some magical way only on the basis of the loaded kotir.

The amazing thing is that he teaches at a month and a half, and tests at 2.5 years. And the test is a good one. I'm afraid there is no peeking ahead.


 
elibrarius:

The amazing thing is that he teaches at a month and a half, and tests at 2.5 years. And the test is a good one.


So what's the result on the white section???? I didn't see anything. All this is very strange, what is so simple. And about neroshel, as far as I know it is no longer supported and not updated, but I can be wrong
1...144114421443144414451446144714481449145014511452145314541455...3495
New comment
Reason: